biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Model] Ultravox Model: Support v0.5 Release (#12912)

Browse Source 
Signed-off-by: Farzad Abdolhosseini <farzad@fixie.ai>
This commit is contained in:
Farzad Abdolhosseini
2025-02-10 14:02:48 -08:00
committed by GitHub
parent 2ae889052c
commit 08b2d845d6
12 changed files with 36 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
tests/distributed/test_pipeline_parallel.py
View File

@@ -215,7 +215,7 @@ MULTIMODAL_MODELS = {
"Qwen/Qwen-VL-Chat": PPTestSettings.fast(trust_remote_code=True),
"Qwen/Qwen2-Audio-7B-Instruct": PPTestSettings.fast(),
"Qwen/Qwen2-VL-2B-Instruct": PPTestSettings.fast(),
"fixie-ai/ultravox-v0_3": PPTestSettings.fast(trust_remote_code=True),
"fixie-ai/ultravox-v0_5-llama-3_2-1b": PPTestSettings.fast(trust_remote_code=True), # noqa: E501
# [Encoder-decoder]
# TODO: Implement PP
# "meta-llama/Llama-3.2-11B-Vision-Instruct": PPTestSettings.fast(),
@@ -234,7 +234,7 @@ TEST_MODELS = [
# [MULTIMODAL GENERATION]
"OpenGVLab/InternVL2-1B",
"microsoft/Phi-3-vision-128k-instruct",
"fixie-ai/ultravox-v0_3",
"fixie-ai/ultravox-v0_5-llama-3_2-1b",
# [LANGUAGE GENERATION - HYBRID ARCH]
"ai21labs/Jamba-tiny-dev",
]

2
tests/entrypoints/openai/test_audio.py
View File

@@ -11,7 +11,7 @@ from vllm.multimodal.utils import encode_audio_base64, fetch_audio
from ...utils import RemoteOpenAIServer
MODEL_NAME = "fixie-ai/ultravox-v0_3"
MODEL_NAME = "fixie-ai/ultravox-v0_5-llama-3_2-1b"
TEST_AUDIO_URLS = [
AudioAsset("winning_call").url,
]

2
tests/entrypoints/test_chat_utils.py
View File

@@ -21,7 +21,7 @@ from ..utils import VLLM_PATH
EXAMPLES_DIR = VLLM_PATH / "examples"
PHI3V_MODEL_ID = "microsoft/Phi-3.5-vision-instruct"
ULTRAVOX_MODEL_ID = "fixie-ai/ultravox-v0_3"
ULTRAVOX_MODEL_ID = "fixie-ai/ultravox-v0_5-llama-3_2-1b"
QWEN2VL_MODEL_ID = "Qwen/Qwen2-VL-2B-Instruct"
MLLAMA_MODEL_ID = "meta-llama/Llama-3.2-11B-Vision-Instruct"
LLAMA_GUARD_MODEL_ID = "meta-llama/Llama-Guard-3-1B"

2
tests/models/decoder_only/audio_language/test_ultravox.py
View File

@@ -15,7 +15,7 @@ from ....conftest import HfRunner, VllmRunner
from ....utils import RemoteOpenAIServer
from ...utils import check_logprobs_close
MODEL_NAME = "fixie-ai/ultravox-v0_3"
MODEL_NAME = "fixie-ai/ultravox-v0_5-llama-3_2-1b"
AudioTuple = Tuple[np.ndarray, int]

2
tests/models/multimodal/processing/test_common.py
View File

@@ -164,7 +164,7 @@ def _test_processing_correctness(
"Qwen/Qwen2-VL-2B-Instruct",
"Qwen/Qwen2.5-VL-3B-Instruct",
"Qwen/Qwen2-Audio-7B-Instruct",
"fixie-ai/ultravox-v0_3",
"fixie-ai/ultravox-v0_5-llama-3_2-1b",
])
@pytest.mark.parametrize("hit_rate", [0.3, 0.5, 1.0])
@pytest.mark.parametrize("num_batches", [32])

2
tests/models/registry.py
View File

@@ -267,7 +267,7 @@ _MULTIMODAL_EXAMPLE_MODELS = {
"Qwen2VLForConditionalGeneration": _HfExamplesInfo("Qwen/Qwen2-VL-2B-Instruct"), # noqa: E501
"Qwen2_5_VLForConditionalGeneration": _HfExamplesInfo("Qwen/Qwen2.5-VL-3B-Instruct", # noqa: E501
min_transformers_version="4.49"), # noqa: E501
"UltravoxModel": _HfExamplesInfo("fixie-ai/ultravox-v0_3",
"UltravoxModel": _HfExamplesInfo("fixie-ai/ultravox-v0_5-llama-3_2-1b",
trust_remote_code=True),
# [Encoder-decoder]
"MllamaForConditionalGeneration": _HfExamplesInfo("meta-llama/Llama-3.2-11B-Vision-Instruct"), # noqa: E501